Derivatization of Surface-Bound Peptides for Mass Spectrometric

Anal. Chem. 2004, 76, 4267-4270

Derivatization of Surface-Bound Peptides for Mass Spectrometric Detection via Threshold Single Photon Ionization Praneeth D. Edirisinghe,† Syed S. Lateef,† Carrie A. Crot,† Luke Hanley,*,† Michael J. Pellin,‡ Wallis F. Calaway,‡ and Jerry F. Moore*,‡

Department of Chemistry, University of Illinois at Chicago, Chicago, Illinois 60607, and Materials Science Division, Argonne National Laboratory, Argonne, Illinois 60439

Chemical derivatization of peptides allows efficient F2 laser single photon ionization (SPI) of Fmoc-derivatized peptides covalently bound to surfaces. Laser desorption photoionization mass spectrometry using 337-nm pulses for desorption and 157.6-nm pulses for threshold SPI forms large ions identified as common peptide fragments bound to either Fmoc or the surface linker. Electronic structure calculations indicate the Fmoc label is behaving as an ionization tag for the entire peptide, lowering the ionization potential of the complex below the 7.87-eV photon energy. This method should allow detection of many molecular species covalently or electrostatically bound to surfaces.

Surface-bound organic and biomolecular species have been developed for a wide range of biochemical applications. Selfassembled monolayers are applied directly or as templates for constructs designed to guide cell growth.1 Arrays of peptides, carbohydrates,2 and proteins3 are utilized in chemical sensing, combinatorial screening, and diagnostics. However, analysis of such complex surface molecular systems has proven challenging. Mass spectrometry is a powerful method for analyzing surfacebound species and for molecular imaging of tissue. Nevertheless, mass spectrometric surface analyses frequently suffer from difficulties in forming ions that are structurally and quantitatively representative of complex organic surface species, especially for multicomponent surfaces. A versatile strategy is demonstrated here to address these shortcomings: threshold single photon ionization (SPI) of desorbed neutral molecules that proceeds by localized ionization of a chemical tag bound to a molecular analyte. Secondary ion mass spectrometry, direct laser desorption, and matrix-assisted laser desorption/ionization (MALDI) use kiloelectronvolt primary ions or pulsed lasers to desorb and ionize organic * To whom correspondence should be addressed: (e-mail) [email protected]. [email protected]. † University of Illinois at Chicago. ‡ Argonne National Laboratory. (1) Chen, C. S.; Mrksich, M.; Huang, S.; Whitesides, G. M.; Ingber, D. E. Science 1997, 276. (2) Ratner, D. M.; Adams, E. W.; Su, J.; O’Keefe, B. R.; Mrksich, M.; Seeberger, P. H. ChemBioChem 2004, 5, 379. (3) Lee, K.-B.; Park, S.-J.; Mirkin, C. A.; Smith, J. C.; Mrksich, M. Science 2002, 295, 1702. 10.1021/ac049434t CCC: $27.50 Published on Web 06/24/2004

© 2004 American Chemical Society

surface species.4-9 These desorption methods have proven extremely powerful for a wide variety of analyses, but they do not utilize the secondary neutrals that are formed with up to 4 orders of magnitude higher abundances than the secondary ions. Furthermore, the number of desorbed ions that do form by energetic desorption methods often fluctuates dramatically due to the competition of different ionization mechanisms and their dependencies upon diverse surface properties. For example, the MALDI signal for a given species is generally found to be inversely proportional to its surface binding strength, inhibiting detection of covalently and other strongly bound species.10,11 Such fluctuations in ion yields complicate identification and quantification of organic surface species, limiting the applicability of these techniques to surface analysis. Photoionization is an attractive approach to improve and normalize signal levels. SPI is a more general detection method than multiphoton ionization5,12-15 due to similar SPI cross sections for many different species. Furthermore, SPI does not require an intermediate state for efficient ionization. Most SPI experiments have been performed by utilizing pulsed5,12,13,16,17 or continuous wave radiation18 with photon energies of g10 eV. Ionization potentials of most organic species range from 7 to 10 eV,19 allowing their ionization by these convenient light sources, but in many (4) Winograd, N. Anal. Chem. 1993, 65, 622. (5) Hanley, L.; Kornienko, O.; Ada, E. T.; Fuoco, E.; Trevor, J. L. J. Mass Spectrom. 1999, 34, 705. (6) Todd, P. J.; Schaaff, T. G.; Chaurand, P.; Caprioli, R. M. J. Mass Spectrom. 2001, 36, 355. (7) Bertrand, P.; Delcorte, A.; Garrison, B. J. Appl. Surf. Sci. 2003, 203/204, 160. (8) Su, J.; Mrksich, M. Langmuir 2003, 19, 4867. (9) Wang, Q.; Jakubowski, J. A.; Sweedler, J. V.; Bohn, P. W. Anal. Chem. 2004, 76, 1. (10) Walker, A. K.; Wu, Y.; Timmons, R. B.; Kinsel, G. R.; Nelson, K. D. Anal. Chem. 1999, 71, 268. (11) Zhang, J.; Kinsel, G. R. Langmuir 2003, 19, 3531. (12) Pallix, J. B.; Schuehle, U.; Becker, C. H.; Huestis, D. L. Anal. Chem. 1989, 61, 805. (13) Van Bramer, S. E.; Johnston, M. V. J. Am. Soc. Mass Spectrom. 1990, 1, 419. (14) McKeown, P. J.; Johnston, M. V. J. Am. Soc. Mass Spectrom. 1991, 2, 103. (15) Koster, C.; Grotemeyer, J. Org. Mass Spectrom. 1992, 27, 463. (16) Becker, C. H.; Wu, K. J. J. Am. Soc. Mass Spectrom. 1995, 6, 883. (17) Trevor, J. L.; Lykke, K. R.; Pellin, M. J.; Hanley, L. Langmuir 1998, 14, 1664. (18) Syage, J. A.; Evans, M. D.; Hanold, K. A. Am. Lab. 2000, 32, 24. (19) King, B. V.; Pellin, M. J.; Moore, J. F.; Veryovkin, I. V.; Tripa, C. E. Appl. Surf. Sci. 2003, 203/204, 244.

Analytical Chemistry, Vol. 76, No. 15, August 1, 2004 4267

cases also depositing up to several electronvolts of excess internal energy into the ions that can lead to fragmentation.13-15 SPI generally induces less fragmentation of molecular species than multiphoton ionization, and this is especially true at photon energies near the ionization threshold. Laboratory sources of tunable vacuum ultraviolet radiation can be used for threshold SPI of organic species, but these sources are not sufficiently intense to saturate SPI,20 a requirement for quantifying the ion signal without independently measuring the photoionization cross section.19 An alternate strategy to achieve threshold SPI is demonstrated here by tuning the ionization potential of the analyte species while using a fixed-wavelength vacuum ultraviolet source, the molecular fluorine (F2) laser. The F2 laser emits at 7.87 eV (157.6 nm) with powers of up to tens of millijoules per pulse, sufficient to saturate SPI but at a photon energy that is too low to achieve SPI of most organic species.19 It is shown here that attaching a chemical tag to an organic species can lower its ionization potential below 7.87 eV to permit SPI with the F2 laser. This strategy will only work if ionization is localized on molecular orbitals of the tag species, as demonstrated here computationally. EXPERIMENTAL SECTION Peptides were synthesized by solid-phase Fmoc synthesis methods, except that the Fmoc group was not removed from the terminal peptide. These Fmoc peptides were then covalently bound to oxidized Si(100) wafers via maleimide coupling to the cysteine residue through an aminopropylsiloxane layer.21,22 Longer peptides bound by these techniques to the chemically similar oxide-coated titanium displayed surface densities of ∼100 pmol/ cm2. Laser desorption single photon ionization (LDPI) was performed by focusing a pulsed nitrogen laser (337 nm) to a spot of a few micrometers diameter using a Schwarzchild objective, an instrument that has been described in detail previously.23-25 Within microseconds of each desorption pulse, a plume of desorbed molecules expanded to fill an ∼50-mm3 volume above the sample (from ion optical calculations). The F2 laser was then fired, ionizing the majority of the desorbed neutrals. Unique electrodes that extract and accelerate the photoions from the large photoionization volume were then pulsed. The ions were refocused with a gridless reflectron and postaccelerated for time-of-flight detection. Each spectrum is the sum of 512 laser shots, each desorbed from a new spot on the surface, and data are plotted as absolute signal without normalization. RESULTS AND DISCUSSION Figure 1 displays the LDPI mass spectra of unlabeled dipeptide Arg-Cys (bottom line) as well as the three Fmoc-derivatized (20) Lipson, R. H.; Shi, Y. J. Ultraviolet and vacuum ultraviolet laser spectroscopy using fluorescence and time-of-flight mass detection. In Ultraviolet Spectroscopy and UV Lasers; Misra, P., Dubinskii, M. A., Eds.; Marcel Dekker: New York, 2002; p 131. (21) Xiao, S.-J.; Textor, M.; Spencer, N. D.; Sigrist, H. Langmuir 1998, 14, 5507. (22) Zreiqat, H.; Akin, F. A.; Howlett, R.; Markovic, B.; Haynes, D.; Lateef, S. S.; Hanley, L. J. Biomed. Mater. Res. 2003, 64A, 105. (23) King, B. V.; Savina, M. R.; Tripa, C. E.; Calaway, W. F.; Veryovkin, I. V.; Moore, J. F.; Pellin, M. J. Nucl. Instrum. Methods Phys. Res. B 2002, 190, 203. (24) Veryovkin, I. V.; Calaway, W. F.; Pellin, M. J. Nucl. Instrum. Methods Phys. Res. A 2004, 519, 353. (25) Veryovkin, I. V.; Chen, C.-Y.; Calaway, W. F.; Pellin, M. J.; Lee, T. Nucl. Instrum. Methods Phys. Res. A 2004, 519, 345.

4268 Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

Figure 1. Laser desorption photoionization using 337- and 157.6nm radiation, respectively, of four peptides covalently bound to a Si surface by maleimide and siloxane linkage chemistry: unlabeled ArgCys and derivatized Fmoc-Arg-Cys, Fmoc-Glu-Cys, and Fmoc-GluTry-Asp-Cys. Ion signal in arbitrary units, with Arg-Cys and other spectra expanded above m/z 200 by ×20 and ×5, respectively.

peptides (from the bottom reading up) Fmoc-Arg-Cys, Fmoc-GluCys, and Fmoc-Glu-Try-Asp-Cys. LDPI of the unlabeled Arg-Cys generated no ion signal above m/z 130 while that of Fmoc-ArgCys showed strong ion signals up to m/z 399. Also observed were two characteristic Fmoc ions at m/z 165 and 179 (see Table 1 for structures). All of these ions were formed by SPI of the Fmoc group followed by fragmentation of the Fmoc-containing parent ion. This hypothesis was supported by the LDPI of physisorbed tyrosine, glycine, and phenylalanine, which only displayed ions above m/z 200 when derivatized with Fmoc (data not shown). LDPI of physisorbed tryptophan also displayed ions above m/z 200 only when derivatized with Fmoc (data not shown), but the IP of tryptophan is e7.5 eV,26 so the underivatized tryptophan must fragment during the desorption or ionization event. High-mass and Fmoc-related ions were similarly observed for covalently bound Fmoc-Glu-Cys and Fmoc-Glu-Try-Asp-Cys, both also shown in Figure 1. Most of the Fmoc-derivatized peptide ions larger than m/z 200 are assigned in Table 1 as deriving from common peptide fragment ions (designated using the modified Roepstorff notation27,28) bound to the Fmoc, maleimide, or maleimide amine functionalities at the expected terminus of the peptide based upon their initial structure. The only exception is the m/z 399 ion of Fmoc-Arg-Cys, which apparently rearranged following photoionization. Finally, m/z values were only accurate to (2, and these mass-derived peak assignments may not correctly reflect the extent of protonation. Previous work found that 355-nm radiation cleaves the C-Si surface bond of silicon surface-bound silanes,5 and a similar cleavage of the aminopropylsiloxane surface bond is expected here. Some of the ions that appear in Figure 1 above m/z 200 have not been identified. These likely result from impurities in the peptides or novel fragmentation pathways during LDPI. The FmocGlu-Try-Asp-Cys spectrum displays the most unidentified ions, (26) Lias, S. G.; Bartmess, J. E.; Liebman, J. F.; Holmes, J. L.; Levin, R. D.; Mallard, W. G. J. Phys. Chem. Ref. Data 1988, 17, 18. (27) Roepstorff, P.; Fohlman, J. Biomed. Mass Spectrom. 1984, 11, 601. (28) Biemann, K. Biomed. Environ. Mass Spectrom. 1988, 16, 99.

Table 1. Peak Assignments for Peaks from LDPI MS of Covalently Bound, Fmoc-Derivatized Peptides (See Figure 1)

which might also result from direct SPI of the tryptophan residue in addition to the Fmoc moiety. Fragmentation pathways during SPI are the subject of ongoing study. Electronic structure calculations were applied to a model Fmocdipeptide to test the hypothesis that threshold SPI at 7.87 eV occurred via ionization of the Fmoc moiety. The ground-state geometric structure of Fmoc-AC is shown in Figure 2, calculated by Gaussian 98 (B3LYP/6-31G).29 The highest occupied molecular orbital is shown to be the conjugated π-orbitals on the Fmoc group. The radical cation [Fmoc-AC]+ is treated similarly, assuming vertical ionization (no geometry relaxation) occurs within a few femtoseconds. The calculated total energy difference between the ground state of the neutral and the ion leads to an ionization potential of 7.64 eV, indicating that 7.87-eV SPI is ∼0.23 eV from threshold. The ionization potential of isolated Fmoc is 7.85 eV while that of the AC dipeptide is even higher.26 Figure 3 displays the variation in net charge per atom for Fmoc-AC upon ionization and indicates that most of the single positive charge of the radical cation is delocalized across the Fmoc group. Additional positive charge is also localized on the first oxygen adjacent to the Fmoc and on the sulfur atom, presumably localized on their lone pair electron orbitals. The results of these calculations clearly show that threshold SPI is feasible in this case and that the resultant net charge is mostly localized on the Fmoc group, which behaves (29) Frisch, M. J.; et al. Gaussian 98; A.9 ed.; Gaussian, Inc.: Pittsburgh, 1998.

as an ionization tag for the entire peptide. A similar effect is expected for the larger peptides measured here. CONCLUSIONS These results have significant implications for the mass spectrometric identification of complex molecular species covalently or electrostatically bound to surfaces. Threshold SPI with 7.87-eV photons can be used to detect many organic species via bound Fmoc, which localizes the ionization event. Furthermore, this effect is not expected to be confined to the Fmoc tag and likely extends to other species such as many of those used as fluorescence probes in optical microscopy30,31 and recently applied to derivatizing surface-bound species.32 The same properties that render a molecular species an efficient fluorescence probe also render it an efficient SPI tag: a highest occupied molecular orbital (30) Craig, D. B.; Dovichi, N. J. Anal. Chem. 1998, 70, 2493. (31) Haugland, R. P. Handbook of Molecular Probes and Research Products, 9th ed.; Molecular Probes: Eugene, OR, 2004. (32) McArthur, E. A.; Ye, T.; Cross, J. P.; Petoud, S.; Borguet, E. J. Am. Chem. Soc. 2004, 126, 2260. (33) Ayre, C. R.; Moro, L.; Becker, C. H. Anal. Chem. 1994, 66, 1610. (34) Aubagnac, J.-L.; Enjalbal, C.; Subra, G.; Bray, A. M.; Combarieu, R.; Martinez, J. J. Mass Spectrom. 1998, 33, 1094. (35) Xu, J.; Szakal, C. W.; Martin, S. E.; Peterson, B. R.; Wucher, A.; Winograd, N. J. Am. Chem. Soc. 2004, 126, 3902. (36) Postawa, Z.; Czerwinski, B.; Szewczyk, M.; Smiley, E. J.; Winograd, N.; Garrison, B. J. Anal. Chem. 2003, 75, 4402. (37) Weibel, D.; Wong, S.; Lockyer, N.; Blenkinsopp, P.; Hill, R.; Vickerman, J. C. Anal. Chem. 2003, 75, 1754.

Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

4269

Figure 3. Variation in charge per atom between Fmoc-Arg-Cys neutral and cation upon Franck-Condon ionization (geometry not reoptimized), calculated by Gaussian 98. Atoms 1-27 correspond to Fmoc group, atoms 28-36 to the alanine (Arg) residue, and atoms 37-51 to the cysteine (Cys) residue.

Figure 2. Optimized geometry and highest occupied molecular orbitals of neutral Fmoc-Arg-Cys-derivatized peptide, calculated by Gaussian 98 (BLYP/6-31G). The vertical ionization potential calculated to be 7.64 eV.

with extended π-conjugation, a relatively low ionization potential, and an ability to stabilize the net positive charge of the radical cation. Threshold ionization leaves little excess energy in the radical cation to induce fragmentation, and the charge stabilization on the tag species may serve to reduce fragmentation, which is often initiated at charge sites. Fragmentation can also be induced by internal energy deposited in the parent ion by the desorption event.5,33 The overall effect of the chemical derivatization/ threshold SPI approach is the appearance of large, clearly identifiable ions from LDPI of these covalently bound peptides. By contrast, the common method of >10-keV Ga+ SIMS of covalently bound peptidesseven those bound with Fmocsoften show only low-mass fragments.34,35 While polyatomic primary ion SIMS has recently been shown effective for the analysis of physisorbed peptides on surfaces (see below),35 the authors are unaware of any such data on covalently bound peptides. Furthermore, the minimum experimental apparatus required to perform these experiments is straightforwardsessentially a MALDI mass spectrometer to which an F2 laser and appropriate delay electronics have been added. Quantification in mass spectrometry is often elusive, especially for desorption from solid surfaces where fluctuations occur in the

4270 Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

efficiency of several competing ionization events.5 However, quantification of organic surface species should also be feasible by chemical derivatization and LDPI. The output of an F2 laser is at least 3 orders of magnitude more intense than other laboratory vacuum ultraviolet sources20 and is sufficient to saturate SPI over a significant volume. The desorption laser can be readily confined to a well-defined surface area, and the chemical derivatization efficiency can be known. Given the capability to saturate the SPI process, all that is required for quantification is measurement of the SPI power versus ion signal curve and calibration of the mass spectrometer transmission. Finally, threshold SPI can be applied to any energetic desorption method including polyatomic primary ion SIMS recently shown to generate large representative ions from various complex organic surface species.5,35-37 ACKNOWLEDGMENT This work is supported by the U.S. Department of Energy, BES-Material Science, under Contract W-31-109-ENG-38 and by the National Institutes of Health (HL64956). The authors acknowledge the assistance of Devin Sears in performing the Gaussian calculations.

Received for review April 14, 2004. Accepted June 8, 2004. AC049434T